Clay Types Modulate the Toxicity of Low Concentrated Copper Oxide Nanoparticles Toward Springtails in Artificial Test Soils

Abstract

Copper oxide nanoparticles (CuO‐NPs) can be applied as an efficient alternative to conventional Cu in agriculture. Negative effects of CuO‐NPs on soil organisms were found, but only in clay‐rich loamy soils. It is hypothesized that clay–NP interactions are the origin of the observed toxic effects. In the present study, artificial Organisation for Economic Co‐operation and Development soils containing 30% of kaolin or montmorillonite as clay type were spiked with 1–32 mg Cu/kg of uncoated CuO‐NPs or CuCl₂. We performed 28‐day reproduction tests with springtails of the species Folsomia candida and recorded the survival, reproduction, dry weight, and Cu content of adults. In a second experiment, molting frequency and the Cu content of exuviae, as well as the biochemical endpoints metallothionein and catalase (CAT) in springtails, were investigated. In the reproduction assay, negative effects on all endpoints were observed, but only in soils containing montmorillonite and mostly for CuO‐NPs. For the biochemical endpoints and Cu content of exuviae, effects were clearly distinct between Cu forms in montmorillonite soil, but a significant reduction compared to the control was only found for CAT activity. Therefore, the reduced CAT activity in CuO‐NP‐montmorillonite soil might be responsible for the observed toxicity, potentially resulting from reactive oxygen species formation overloading the antioxidant system. This process seems to be highly concentration‐dependent, because all endpoints investigated in reproduction and biochemical assays of CuO‐NP‐montmorillonite treatments showed a nonlinear dose–response relationship and were constantly reduced by approximately 40% at a field‐realistic concentration of 3 mg/kg, but not at 32 mg/kg. The results underline that clay–CuO‐NP interactions are crucial for their toxic behavior, especially at low, field‐realistic concentrations, which should be considered for risk assessment of CuO‐NPs. Environ Toxicol Chem 2022;41:2454–2465. © 2022 The Authors. Environmental Toxicology and Chemistry published by Wiley Periodicals LLC on behalf of SETAC.

Biomarkers, Copper, Soil ecotoxicology, Soil invertebrates, Nanoparticles

INTRODUCTION

Copper‐based nanoparticles (CuO‐NPs) are already in use as an efficient alternative to conventional Cu formulations in agriculture, with two main manners of application: as a Cu source in fertilizers (Adisa et al., 2019; Dimkpa & Bindraban, 2018) or as bactericides and fungicides (Kah et al., 2019; Keller et al., 2017; Peixoto et al., 2021; Su et al., 2020). These particles can be more efficient than conventional, for example due to a higher toxicity toward target organisms (Rai et al., 2018) or reduced wash‐off from crop surfaces (Kah et al., 2019). Although they are not directly applied in the field, CuO‐NPs are widely tested as a model substance for their toxicity to plants (Gao et al., 2018; McManus et al., 2018; Stewart et al., 2015), freshwater organisms (Heinlaan et al., 2008; Janani et al., 2020; Manusadžianas et al., 2012), marine organisms (Buffet et al., 2011; T. Gomes et al., 2011; Pang et al., 2013), soil organisms (Bicho et al., 2017; Mendes et al., 2018; Noordhoek et al., 2018), and microorganisms (Dimkpa et al., 2011; Khalid et al., 2021; Kim et al., 2013), which was also reviewed by Hou et al. (2017).

In these studies, toxic effects tended to be lowest when organisms were exposed in soils, probably due to the binding capacity of the soil matrix for Cu (Jarvis, 1981; Wu et al., 1999). However, CuO‐NPs can also be toxic to soil invertebrates and plants, but only in clay‐rich loamy soils, indicating a crucial interaction of CuO‐NPs with clay minerals (Fischer et al., 2021a; Yusefi‐Tanha et al., 2020). These soils are beneficial to plant growth and therefore of great importance for agriculture (Arora et al., 2011; Dexter, 2004; Jalota et al., 2010; Tracy et al., 2013). In contrast, most studies reporting no or only slight toxic effects of CuO‐NPs on soil invertebrates were conducted in sandy standard soils (e.g., Bicho et al., 2017, 2020; Mendes et al., 2018; Noordhoek et al., 2018), where the lack of effects could be due to the lower availability of clay minerals for CuO–NP interactions.

The composition of a soil clay fraction is highly diverse, but the components can roughly be classified into two‐ and three‐layer clay minerals. The latter have the ability to widen the distance between their mineral layers by incorporating water, resulting in a higher specific surface area and cation exchange capacity (CEC). Because of these properties, a high clay content in soils is usually expected to reduce the bioavailability and toxicity of metals in soils because it allows them to strongly adsorb metals and metal‐based NPs (Abbas et al., 2020; Golia et al., 2008; Lanno et al., 2019; Sungur et al., 2014). However, in some cases clays may also increase the toxicity of metals, for example Cu and clays can form highly reactive associations (Kalidhasan et al., 2017; Khanikar & Bhattacharyya, 2013) inducing strong toxicity to microbial cells (Bagchi et al., 2013; Das et al., 2014; Hundáková et al., 2013; Malachová et al., 2011; Pourabolghasem et al., 2016; Sohrabnezhad et al., 2014). Furthermore, for some metal‐based NPs an increased toxicity to protozoans (Gupta et al., 2017) and zebrafish was observed when applied together with clays (Gupta et al., 2016). In all of these studies, three‐layer clay minerals, mainly montmorillonite, were used.

Springtails (Collembola) have multiple ways to react toward metal contamination in soils. In the case of Cu, its role as an essential element has to be considered because this characteristic strongly drives metal kinetics in soil invertebrates (Ardestani et al., 2014). Small amounts of Cu can even be beneficial to springtails (Pfeffer et al., 2010). However, increasing soil Cu concentrations causes an up‐regulation of Cu excretion mechanisms in an attempt to maintain a bearable Cu body content (Fountain & Hopkin, 2001; Pedersen et al., 2000; van Straalen et al., 1987). One of these mechanisms is molting, which enables the removal of exuviae but also of the midgut epithelia, where metals can be bound for detoxification (Fountain & Hopkin, 2001; van Straalen & Roelofs, 2005). This is often accompanied by and linked with the upregulation of metal‐adsorbing proteins, metallothioneins (MTs; Maria et al., 2014; Nota et al., 2011; Xiong et al., 2014). However, detoxification mechanisms are energy‐consuming processes (Jager et al., 2013) that may reduce Cu body concentration, but also impact reproduction and growth (Bednarska et al., 2013). This may explain previous observations where toxic effects of CuO‐NPs on reproduction and growth co‐occurred with low Cu body concentrations in springtails (Fischer et al., 2021a).

Beside excretion, antioxidative enzymes such as catalase (CAT) can be activated as a defence mechanism against the oxidative stress induced by Cu salts (Maria et al., 2014; Xiong et al., 2014) and nanosized Cu in the soil (Gautam et al., 2018; S. I. Gomes et al., 2011). Copper oxide nanoparticles are retained in the epithelial gut cells of invertebrates (Heinlaan et al., 2011; Velicogna et al., 2021) and are able to induce the formation of reactive oxygen species (ROS) directly in these cells (Baeg et al., 2018), which can be broken down by the CAT enzyme.

In the present study, we exposed the standard test organism Folsomia candida to CuO‐NPs and CuCl₂ in Organisation for Economic Co‐operation and Development (OECD) soils containing either kaolin or montmorillonite as the clay fraction. Both clays are either exemplary for two‐ or three‐layer clay minerals and are widely used, for example kaolin in OECD soil and Montmorillonite K10 in different catalytic (Kalidhasan et al., 2017; Khanikar & Bhattacharyya, 2013) and ecotoxicological studies (Gupta et al., 2016, 2017; Kansara et al., 2019). We compared the impact of these soils on the toxicity of the Cu substances to the survival, reproduction, growth, molting, and biochemical responses of F. candida. We hypothesized that soils containing montmorillonite clay would affect springtails more strongly by forming more reactive Cu–clay associations than soils containing kaolin clay.

MATERIALS AND METHODS

Test soils

Artificial OECD test soils differing in clay content and composition were produced according to OECD Guideline 232 (OECD, 2009) with some modifications. The organic carbon content (C_org), pH, and clay content were aligned to previously tested loamy field soils (Refesol 05‐G; Fischer et al., 2021a). The soils contained two different clay fractions consisting either of kaolin (Erbslöh Lohrheim GmbH KG), as suggested by OECD guidelines (OECD, 2009), or chemically activated Montmorillonite K10 (CAS‐No. 1318‐93‐0; Sigma‐Aldrich). The final soils contained either of the two clay types at 19% or 30% of the total solid fraction, resulting in four different exposure media (Table 1). The reproduction assay was conducted for all four soils, but the results for the soils containing 19% of either kaolin (K19) or montmorillonite (M19) are shown in Supporting Information, Figures S1 and S2 because effects were much less pronounced than in soils containing 30% of either kaolin (K30) or montmorillonite (M30).

Table 1:

Properties of test soils

Test soil	Clay fraction	Sand in mass %	Clay in mass %	Corg in mass %	pH (CaCl2)	WHC In g/kg	Cu in mg/kg	Fe in g/kg
K19a	Kaolin	81	19	1.59	5.15	304	3.4	1.45
K30	Kaolin	70	30	1.59	5.17	360	6.4	1.45
M19a	MMT	81	19	1.59	5.10	410	0.31	5.57
M30	MMT	70	30	1.59	5.16	457	0.36	6.96

a

Data shown in Supporting Information, Figures S1 and S2.

C_org = organic carbon content; MMT = montmorillonite K10; WHC = water holding capacity.

Peat (Einheitserdewerke Patzer GmbH KG) was air‐dried and finely ground with a blender to achieve the maximum possible homogenous distribution of C_org in the soils. The pH of the test soils was measured (inoLab pH Level 1; WTW) in CaCl₂ and adjusted with CaCO₃ to 5.2 ± 0.1 and the water holding capacity of freshly mixed soils was measured. The Cu content of the test soils was analyzed by graphite furnace atomic absorption spectroscopy (AAS; see Statistics) after acid extraction with 65% HNO₃ for 2 h at 80 °C (Fischer et al., 2021a; Pedersen & van Gestel, 2001) and the Fe content was estimated by energy dispersive X‐ray fluorescence (ED‐XRF) spectroscopy using a PANalytical Epsilon 3‐XL benchtop ED‐XRF spectrometer (Malvern Instruments; Wien et al., 2005).

Test substance and soil spiking

Copper oxide nanoparticles were purchased from Sigma‐Aldrich. According to the supplier, they have a spherical shape, a particle diameter of <50 nm, a surface area of 29 m²/g, and a crystalline phase of tenorite, which is stated by many studies and summarized by Fischer et al. (2021b). The hydrodynamic diameter (non‐negative least squares) and zeta potential between 0 and 7 days in miliQ water are 500–1000 nm and +8 to +11 mV (Fischer et al., 2021b). For each test, a fresh stock dispersion of CuO‐NPs in miliQ water at a concentration of 300–600 mg Cu/L was prepared and ultrasonicated for 30 min at 35 kHz (Sonorex RE100H; Bandelin Electronic GmbH KG) directly before application. For Cu salt treatments, a stock solution of CuCl₂ × 2H₂O (Merck KGaA) at a concentration of 1.000 mg/L was prepared and used for all assays. The respective stock solutions and dispersions were diluted in miliQ water to the required concentrations to achieve soil Cu concentrations of 1, 3, 10, and 32 mg/kg soil dry weight above the background and directly pipetted on and stirred into the soil until homogenous distribution was achieved. Soil moistened with miliQ water only served as negative control. The water content of all soils was set to 50% of their water holding capacity (WHC). Soils were stored at 17 °C in the dark for 2–3 days and water content was adjusted directly before use in the bioassays.

Bioassays

All F. candida adults originated from our laboratory cultures which are maintained on charcoal‐plaster plates at 15 °C in the dark. Synchronizations were conducted at room temperature. In all bioassays, juveniles were 10–12 days old at the start of the test. All test vessels were stored at 20 °C under a 16:8‐h light:dark cycle during the test and provided with a small amount of dry yeast at test start and at least once a week until test end.

Reproduction assay

The reproduction test was conducted according to a miniaturized protocol (Filser et al., 2014) which is based on OECD Guideline 232 (OECD, 2009). In brief, cohorts of five F. candida juveniles were exposed to 10 g of moist soil (50% WHC) in a 30 ml snap‐lid glass jar (VWR) in five replicates per concentration. Once a week, the jars were aerated and dry yeast was added as food source. The water content was controlled at the beginning and the end of the test by weighing the test vessels. After 4 weeks, the soil from the test vessels was poured into plastic beakers, and the animals were floated by adding 100 ml of tap water stained with a dash of black ink and stirred gently. The surface of the beakers was photographed (Panasonic Lumix DMC‐FZ50) and the number of adults and juveniles was counted manually with the software ImageJ (Schneider et al., 2017). Adults were collected from the water surface with a small steel spatula and transferred onto clean culture plates of charcoal plaster for 1 h. Afterwards, they were collected in an Eppendorf cup, deep‐frozen at −20 °C and freeze‐dried (alpha 1‐2 LDplus; Christ). The Eppendorf cups containing the freeze‐dried springtails were weighed in an ultra‐fine balance (d = 0.01 mg; Sartorius BP 211 D). The previously assessed weights of the cups were subtracted and the results divided by the number of collected animals to end up with the average individual dry weight. The freeze‐dried springtail samples were further used for AAS analysis. Tests were considered valid according to the OECD Guideline (OECD, 2009) when in the control the survival of adults was ≥80% and on average more than 100 juveniles per vessel were produced.

Molting and biochemical assays

For the molting and the biochemical experiments, all animals were exposed to 30 g of moist soil in 200‐ml plastic beakers for 6 days. The animals again were extracted by floating with tap water and ink as described for the reproduction assay. For the molting experiments, 30 animals were transferred to clean culture plates and kept in a climate chamber for 1 week. After 2, 5, and 7 days, exuviae shed on the plates were counted and collected. Due to the high variation in molting behavior, these experiments were conducted twice with five replicates each.

For the CAT and MT measurements, 50 and 150 animals, respectively, were transferred to culture plates after the 6 days of exposure. Afterwards, the animals were collected in Eppendorf cups, deep frozen in liquid nitrogen and stored at −80 °C until further analysis. The MT assay was conducted according to Maria et al. (2014) and Viarengo et al. (1997) with slight adaptions. Briefly, the animals were manually homogenized with a micropestle and ultrasonicated at 35 kHz for 1 min in 600 µl of 0.1 mM potassium phosphate buffer (pH = 7.4) containing 1 mM ethylenediaminetetraacetic acid (EDTA; Sigma‐Aldrich) and 1 mM dithiothreitol. To guarantee complete dissolution of EDTA, the buffer was first ultrasonicated for 5 min at 30 °C. Then, 500 µl of the homogenate were mixed with 500 µl of an EtOH/chloroform/miliQ water solution (87/8/5% v/v) and centrifuged at 6000g (10 min, 4 °C). The ethanol had been cooled to −20 °C before use. From the supernatant, 750 µl was mixed with 1.2 ml of EtOH and 50 µl of an RNA dispersion (20 g/l) and frozen at −20 °C for 2 h. After centrifuging at 6000g (1 min, 4 °C), the supernatant was discarded, and the precipitate was resuspended in 300 µl of an EtOH/chloroform/miliQ water solution (87/1/12% v/v). All samples were centrifuged again at 6000g (1 min, 4 °C). The supernatant was discarded, and the precipitate was resuspended again in 150 µl of 0.25 M NaCl, 150 µl of Tris‐EDTA buffer (5 and 4 mM, respectively, pH = 7), 300 µl of Ellman's reagent (0.4 mM 5,5‐dithio‐bis‐(2‐nitrobenzoic acid; Sigma‐Aldrich), and 2 M NaCl in 0.2 M KH₂PO₄, pH = 8). As a standard, reduced glutathione (Sigma‐Aldrich) was diluted to different concentrations in the Tris‐EDTA buffer and mixed in a 1:1 ratio with Ellman's reagent. From each sample and standard concentration, 190 µl was transferred to a 96‐well plate (Sarstedt) in triplicate, respectively, and the absorbance was measured after 5 min at 405 nm (Wallac Victor2 1420 Multilabel Counter; Perkin Elmer).

The CAT assay was conducted according to Cohen et al. (1970). In brief, 50 juveniles were manually homogenized in 0.01 M sodium phosphate buffer (pH = 7.4). After centrifugation at 10 000g (4 °C, 10 min), 10 µl of supernatant was transferred in triplicate to a 96‐well plate. Afterwards, 100 µl of 6 mM H₂O₂ was added and after 3 min the degradation of H₂O₂ was interrupted by the addition of 25 µl H₂SO₄ (6 N). Then, 125 µl of 0.02 mM KMnO₄ was added and the absorbance was immediately measured at 490 nm. The CAT activity was calculated based on the absorbance of the samples in relation to the absorbance of a standard containing no H₂O₂.

The MT level and CAT activity per milligram of protein of the used homogenate or supernatant were calculated, respectively. Protein content was assessed according to Bradford (1976) with a microassay (Bio‐Rad) using bovine serum albumin as standard and Bradford reagent as dye. From each sample, 5 µl was transferred in triplicate to a 96‐well plate, 250 µl of Bradford reagent was added, and the absorbance was read at 620 nm after 5 min.

Cu analytics

All springtail and exuviae samples were digested in 125 µl of 65% HNO₃ and 30% H₂O₂, respectively, for 1 h at 65 °C, then for 1 h at 85 °C and, finally, overnight at 95 °C with lids open until complete dryness. The samples were dissolved in 1% HNO₃ and the Cu concentration was measured with AAS using a Uni‐cam989QZ AA Spectrometer (Unicam) with a GF‐90+ furnace and an FS‐90+ autosampler. The Cu recovery of this method was assessed to 78.4% ± 7.1% by digesting pig kidney (ERM‐BB186; European Commission Joint Research Center) as reference material.

Statistics

Statistical analyses were conducted with RStudio (R Core Team, 2018). For the reproduction and biochemical data, all treatments were compared to the control by a generalized linear model using Gaussian or gamma as the distribution family, depending on the best model fit. Data were only transformed if the normal distribution of the model residuals was not given. Comparison of substance groups as done for the biochemical data were also done by a generalized linear model. Due to the two test runs, the data from the molting experiment were fitted by a nested analysis of variance using a linear mixed effect model from the package nlme (Pinheiro et al., 2019) with the test number as random factor; in this case, differences from the control were analyzed by a Dunnett post hoc test. Overall comparison of endpoints in M30 soil was done by a linear mixed effect model where test concentration was nested within endpoints, using the lme4‐package (Bates et al., 2015). In Cu quantification datasets containing <25% of data points below the limit of quantification (LOQ), these data points were substituted by LOQ/sqrt(2; Tekindal et al., 2017). In datasets containing ≥25% of data points below LOQ, statistical analysis was omitted or analyzed by ordinal logistic regression with the MASS package (Ripley et al., 2002).

RESULTS

Survival, reproduction, and dry weight

In the 28‐day reproduction tests, the survival, dry weight, and Cu content of the springtails introduced at the test start were investigated. Negative effects on springtails were mainly found in the montmorillonite soil M30 and only for NP treatments (Figure 1). The survival and reproduction of springtails in the kaolin soil K30 were never impaired by more than 28% of the control, respectively. In M30 soil, survival was reduced most in NP‐1 by −52% (p = 0.002; Figure 1B) and reproduction in NP‐3 by −33% (p = 0.045; Figure 1D). The dry weight of springtails in K30 soil was constantly elevated in both Cu salt (S) and NP treatments, although only significantly in S‐10 (+94%, p = 0.03; Figure 1E). In M30 soil, dry weight was lowered most in NP‐3 by −42% (p = 0.049).

Figure 1:

Survival, reproduction, and dry weight of springtails. Mean effects in % of control ± SE (n = 5) on the survival (A,B), reproduction (C,D) and dry weight of springtails (E,F) exposed to CuO‐nanoparticles (NP) or CuCl2 (S), either exposed in kaolin soil K30 (A,C,E) or montmorillonite soil M30 (B,D,F). Significant differences to the control analyzed by glm with 0.01 < p < 0.05 are marked by asterisks with * = 0.01 < p < 0.05 and ** = 0.001 < p < 0.01.

Cu body concentration

The Cu content of adults was generally much higher in K30 soil (Figure 2A). This can be attributed to the higher background Cu content of the kaolin (Table 1). There was no remarkable difference in Cu content between CuO‐NP and CuCl₂ treatments in all soils. In K30 soil, the control animals contained 34 ± 8 µg Cu/g dry weight (mean ± SE, n = 5). In CuO‐NP treatments, the Cu content increased from 81 ± 31 to 107 ± 22 µg Cu/g dry weight and in CuCl₂ treatments from 52 ± 9 to 118 ± 18 µg Cu/g dry weight by the addition of 1–32 mg Cu/kg, respectively. For M30 soil, all treatments were below the LOQ, except treatments M30‐NP‐32 and M30‐S‐32, where Cu body contents were remarkably elevated to 74 ± 43 and 62 ± 27 µg Cu/g dry weight, respectively.

Figure 2:

Cu body concentration of springtails. Mean concentration ± SE (n = 5) of bioaccumulated Cu in kaolin soil K30 (A) and montmorillonite soil M30 (B). Significant differences to the control analyzed by glm are marked by asterisks with * = 0.01 < p < 0.05, ** = 0.001 < p < 0.01 and *** = p < 0.001. The dashed line indicates the limit of quantification (LOQ) divided by the average dry weight in M30 soil. Because most values from 0–10 mg Cu/kg in M30 soil were <LOQ, no statistical analysis was conducted for that soil.

CAT activity and MT levels

Overall, the CAT activity of springtails was inhibited by Cu in both soils. In the organisms exposed to the M30 soil, overall this effect was felt the most in CuO‐NP treatments rather than CuCl₂ treatments (p = 0.016; Figure 3B). The most pronounced CAT inhibition, −62% as compared to the control (p = 0.012), was observed in NP‐10, followed by −39% (p = 0.093) in NP‐3 of M30 soil. In the organisms exposed in the K30 soil, the CAT activity was significantly inhibited in NP‐1 (−40%, p = 0.013) and S‐10 (−36%, p = 0.023), but the overall difference in inhibition between CuO‐NP and CuCl₂ treatments was not significant (p = 0.91; Figure 3A).

Figure 3:

Catalase activity and metallothionein levels in springtails. Mean effects in % of control ± SE (n = 3) on the catalase activity (A,B) and metallothionein levels (C,D) of springtails exposed to CuO‐nanoparticles (NP) or CuCl2 (S), either exposed in kaolin soil K30 (A,C) or montmorillonite soil M30 (B,D). Significant differences between substances NP and S and of treatments to the control analyzed by glm with 0.01 < p < 0.05 are marked by an asterisk. CAT = catalase; MT = metallothionein.

The MT expression was constantly lowered by Cu in both soils and, as in the case of CAT, the difference between CuO‐NPs and CuCl₂ was only significant in M30 soil (p = 0.035; Figure 3D), but not in K30 soil (p = 0.97; Figure 3C). Except for CuCl₂ in K30 soils, MT levels occurred as U‐shaped curves in relation to Cu concentrations.

Number and Cu content of exuviae

There was no statistically significant difference in the number of collected exuviae between treatments in all soils, but in K30 soil it was significantly increased by 57% (p < 0.001) compared to M30 soil (Supporting Information, Figure S3). The highest increase in the number of exuviae compared to the control was found in M30 soil for NP‐3, the treatment with the most observed negative effect on reproduction. In this treatment, molting was increased by 52% within 3–5 days after exposure, although not significantly (Supporting Information, Figure S3C).

The Cu content of the exuviae was in most cases below LOQ. For values above LOQ, the highest exuviae Cu content was recorded in M30 soil for NP‐3 (Figure 4). We calculated the probability of samples in the respective soils and substances (control, S or NP) to exceed the limit of detection or LOQ. The odd ratios of the likelihood to exceed one of these ordinal limits for the respective substance or soil groups are given in Supporting Information, Table S1. On average, the likelihood of exceedance was 35% lower for NP samples and 66% lower for S samples compared to the control. With respect to soil types, this likelihood was approximately five times higher in M30 soil than in K30 soil.

Figure 4:

Cu content of springtail exuviae. n = 10, in pg Cu/skin exposed to CuO‐nanoparticles (NPs) or CuCl2 (S), in kaolin soil K30 (A) or montmorillonite soil M30 (B). Copper contents are based on measured values >limit of quantification (LOQ; blue), <LOQ and >limit of detection (LOD; green) and <LOD (yellow). Substances (control, NP, and S) are optically separated by the dashed lines. Only samples containing >9 exuviae were considered; explanations for this decision are given in Supporting Information, Figures S6 and S7.

DISCUSSION

The nonlinear dose–response relationship

In the present study, reproduction rates in the M30 soil corroborate the observation that, in contrast to CuCl₂, CuO‐NP can negatively impact F. candida in clay‐rich soils in a nonlinear dose–response relationship (Fischer et al., 2021a) at field‐realistic test concentrations that are considered to be in the range of 0.5–20 mg Cu/kg (calculated fresh Cu input within one to five seasons [Fischer et al., 2021a]) or one season of application (Peixoto et al., 2021). The reason for this nonlinear dose–response relationship may be that the stronger agglomeration behavior of CuO‐NPs at test concentrations ≥50 mg Cu/kg (Fischer et al., 2021a; Velicogna et al., 2021) reduces their reactive surfaces. This may also explain why no toxic effects on soil invertebrates are typically observed at much higher concentrations of CuO‐NP (summarized by Fischer et al., 2021a) or other metal‐based NPs in the range of 1000–6400 mg/kg (Heckmann et al., 2011; Noordhoek et al., 2018; Pereira et al., 2011). Consequently, at lower soil concentrations less aggregated NPs may exhibit more reactive surfaces. In lungs, the ROS‐generating capability of NPs is directly related to their surface area, which massively increases with decreasing particle size (Nel et al., 2006). For CeO₂‐NP‐induced DNA damage and oxidative stress in spermatozoans, an inverse dose–response relationship was also explained by a reduced particle agglomeration at low concentrations and a resulting better accessibility to spermatozoans (Cotena et al., 2020; Préaubert et al., 2018).

The U‐shaped CAT activity curve in M30 soil coincides with the observed low‐concentration effects for the endpoints of the reproduction assay, indicating that impaired CAT activity could have reduced the survival, reproduction, and dry weight of springtails in that soil. At 3 mg Cu/kg, all endpoints were approximately 40% lower than the control and the overall significant difference was most pronounced, while no significant effect was visible anymore at 32 mg Cu/kg (Figure 5). The CuO‐NPs tend to agglomerate in soils with increasing NP concentration (Velicogna et al., 2021), with a distinct increase in CuO‐NP particle size from 15 to 50 mg/kg, indicating an enhanced agglomeration (Fischer et al., 2021a). This might explain why toxic effects in M30 soil did not occur anymore at 32 mg/kg. Negative effect peaks of the different endpoints varying in the test concentrations of 1 and 10 mg Cu/kg may be due to trade‐offs in F. candida (Crommentuijn et al., 1997). At 1 mg Cu/kg, possibly only the fittest adults survived the chemical stress caused by CuO‐NPs and, consequently, were able to use all available food and space resources for above‐average growth and reproduction per individual (Figure 5). For F. candida it is known that lower initial population densities allow a higher egg production per individual (Filser et al., 2014), even under chemical stress (Noel et al., 2006). Crommentuijn et al. (1997) observed a trade‐off between growth, reproduction, and survival in response to chemical stress in this species, however, with survival prioritized over growth and reproduction.

Figure 5:

Dose–response relationship of all biological endpoints for copper oxide nanoparticles (CuO‐NP) in M30 soil. Shown as mean effects in % of control for survival, reproduction, dry weight (n = 5, 28 days of exposure) as well as catalase activity and metallothionein level (n = 3, 6 days of exposure) of F. candida exposed to CuO‐NP in M30 soil. Significant differences to the control, analyzed by a linear mixed effect model where test concentration was nested within endpoints, are marked by asterisks: * = 0.01 < p < 0.05, ** = 0.001 < p < 0.01, *** = 0.001 < p. CAT = catalase; MT = metallothionein.

The impact of different clay minerals

The absence of negative effects on survival and reproduction plus the positive effect on dry weight in K30 soil underline how the clay type interacting with Cu influences its toxicity. The Cu uptake via food results in a positive hormetic effect on the reproduction of F. candida at low Cu concentrations (Pfeffer et al., 2010). This may also be the case for Cu–kaolin associations, which seem to support growth. In contrast, the presented data suggest that the combination of CuO‐NPs and montmorillonite can affect springtails on a physiological and a molecular level. We suggest that the ability of this combination of substances to form ROS can explain these effects, as discussed in the following section, Cu contents and detoxification.

Catalase activity has been reported to increase (Buffet et al., 2011; T. Gomes et al., 2012; Maria et al., 2014; Ribeiro et al., 2015) or decrease in invertebrates after exposure to metals and metal NPs (García‐Gómez et al., 2019; S. I. Gomes et al., 2011; Li et al., 2020; Xiong et al., 2014). In some studies on organic and inorganic contaminants, CAT activity has been reported to increase at low concentrations or for short durations, and to diminish at higher concentrations or longer exposure durations (Du et al., 2015; Hu et al., 2010; Lin et al., 2010; Ma et al., 2017; Wang et al., 2014; Zheng et al., 2022). In these studies, the effect of reduced CAT activity was explained by an overload of the antioxidant defence system by the presence of excess ROS, which in turn oversaturated the antioxidant system. Ma et al. (2017) showed that the higher the CAT activity in snails was increased by Cu in the short term, the more it was depleted in the long term. Sustained decreases in CAT activity has been observed in several species following exposure to ionic Cu (Garcia Sampaio et al., 2008; Gautam et al., 2018; Vutukuru et al., 2006), CuO‐NPs (Canli et al., 2017; Janani et al., 2020), and Cu‐NPs (Gautam et al., 2018).

In the present study, the marked CAT inhibition in M30‐NP treatments can be explained by a higher ROS formation overloading the antioxidant system. Reactive oxygen species formation may be caused by the high reactivity of Cu–montmorillonite associations, which are able to break down organic compounds and are known to be more reactive than Cu–kaolin (Khanikar & Bhattacharyya, 2013) or Cu–sand associations (Kalidhasan et al., 2017). For example, Cu–montmorillonite associations can cause a high degradation rate of atrazine through the activation of molecular oxygen and the subsequent hydroxyl radical formation (Hong et al., 2017). The Montmorillonite K10 clay used in the present study is rich in structural and exchangeable Fe (Wirth, 2005), which is also reflected in the total Fe content of our test soils (Table 1). Both types of Fe can act as an electron shuttle to reduce metal ions (Liu et al., 2014). We suggest that metallic Fe associated with Montmorillonite K10, which was found in X‐ray diffraction spectra by Santangelo et al. (2011) and in the present study (Supporting Information, Figure S4), acts as a Lewis base on dissolution, thus fueling the electron shuttle via octahedral Fe(III), (Wirth, 2005) to the CuO‐NPs. Finally, the reduced Cu(I) can provide electrons for ROS formation on the NP surface. The decline in CAT activity only for CuO‐NPs may be due to the fact that Cu²⁺ can be adsorbed to a large amount by interlayer binding sites (Dau & Lagaly, 1998), which is not possible for the CuO‐NPs of ~50 nm used in our study (Pourabolghasem et al., 2016; Sohrabnezhad et al., 2014). Instead, they may stay adsorbed on the clay mineral surface, where they are able to directly interact with their environment.

The addition of Montmorillonite K10 has been found to increase ROS formation in protozoans by ZnO‐NPs in the long term (Gupta et al., 2017), although in zebrafish the association between CuO‐NPs and Montmorillonite K10 has been found to cause the opposite effect (Kansara et al., 2019). However, in the latter study increased sedimentation of CuO‐NPs caused by clay addition may be also responsible for reduced ROS formation.

Moreover, Cu²⁺ ions have the ability to reduce CAT activity by directly interacting with the CAT enzyme and changing its secondary structure (Hao et al., 2015). However, this does not explain why the most pronounced reduction in CAT activity occurred in the soil with exceptionally lower Cu pore water concentration (Supporting Information, Figure S5). We therefore conclude that, in the present case, an inhibition of CAT activity through ROS formation is more likely.

Cu contents and detoxification

Copper body content data give further support to the previous observation that the accumulated Cu quantity cannot explain toxic effects (Fischer et al., 2021a), because these effects were never found in treatments with high Cu body contents of springtails (Figures 1 and 2). In general, the critical body load of trace metals in invertebrates is determined by the metabolically available fraction and not the total metal body load (Rainbow, 2002). Furthermore, detoxification mechanisms can have direct or indirect effects on organisms’ energy budget (Jager et al., 2013), consequently affecting reproduction and growth rates (Bednarska et al., 2013). We assume that Cu excretion mechanisms, which aim to reduce the Cu body concentration, were at play in F. candida and might have contributed to the effects observed in our study.

Metallothionein levels in Cu treatments were constantly lower than in the respective controls, but due to the high variability of the data statistical significance was rarely found. Higher MT levels in the control compared to Cu‐treated soils have also been observed in earthworms exposed to copper sulfate (Xiong et al., 2014). In M30 soil, however, the significant difference between CuO‐NPs and CuCl₂ in MT levels suggests a higher molecular defence reaction of springtails for CuO‐NPs than for Cu²⁺ when bound to montmorillonite. Furthermore, absolute MT levels are approximately twice as high in M30 compared to K30 soil (Supporting Information, Table S2). Overall, the highest absolute MT levels in Cu treatments coincided with the highest observed toxicity and CAT depletion, namely in M30‐NP treatments (Figures 1 and 5). We conclude that the higher MT levels of CuO‐NPs compared to CuCl₂ observed in our study might be a response toward the toxic properties of CuO‐NP–montmorillonite associations.

These higher MT levels are accompanied by higher Cu levels in springtails' exuviae when exposed to CuO‐NPs and/or in M soil (Figure 4). In the case of the latter, it might be due to the comparably stronger attachment of montmorillonite toward cell membranes (Bagchi et al., 2013; Das et al., 2014; Pourabolghasem et al., 2016). In our experiment, collected exuviae could still contain gut residues as they are shed together during the molting process. These gut residues could contain more Cu than the exuviae themselves, as observed for the springtail Orchesella cincta (van Straalen et al., 1987). This is supported by the fact that on the one side the springtail cuticle is highly omniphobic, that is, it repels both hydrophilic and lipophilic substances (Hensel et al., 2016; Kong et al., 2019), and on the other side the main uptake route for chemicals into springtail is the pore water (Ogungbemi & van Gestel, 2018; Ronday et al., 1997). We suggest that CuO‐NPs were attached more to shed exuviae/midguts than CuCl₂, even more when montmorillonite instead of kaolin was involved. Roughly estimated from the number and Cu content of exuviae, molting may have contributed approximately 10%–20% to the difference in the total Cu body content of springtails between M30 and K30 soils after 28 days of exposure (Figure 2). The main reasons for this difference, however, probably are the relatively higher Cu background in K soils (Table 1) and the lower CEC of kaolin for Cu compared to montmorillonite (Farrah & Pickering, 1976).

In the low concentration Cu treatments of M30 soil, the measured Cu body concentrations are clearly below the initial Cu body concentration of F. candida taken from cultures (17–75 µg/g; Ardestani & Van Gestel, 2013; Pedersen et al., 1997) or exposed to unspiked field soils (20–88 µg/g; Ardestani & Van Gestel, 2013; Bongers, 2007; Pedersen et al., 2000; Santorufo et al., 2012). However, the Cu content of F. candida from our culture reared for 6 weeks on plaster of Paris was also below the LOQ of AAS. A very low Cu body content of F. candida exposed to plaster of Paris for longer was also observed in other studies (Fountain & Hopkin, 2001; Pedersen et al., 2000). The low Cu body content is probably due to the high adsorption ability of plaster of Paris, which potentially reduces Cu uptake by springtails. We conclude that the same mechanism can be expected for M30 soil, which contains highly Cu adsorbing montmorillonite (Farrah & Pickering, 1976) and thus can reduce the springtail Cu body concentration in the long term by Cu deprivation.

CONCLUSION

The present study indicates that montmorillonite as a reactive three‐layer clay mineral can increase the toxicity of CuO‐NPs to springtails. The highly reduced CAT activity in springtails indicates that this toxicity is most likely mediated through ROS formation by CuO‐NP–montmorillonite associations. According to our U‐shaped data generated for CuO‐NPs in soil containing montmorillonite clay (Figure 5), this process is highly concentration‐dependent, and deviating effect peaks of the different endpoints might be due to trade‐offs from one trait to another in springtails. Moreover, deleterious effects were highest in the range of field‐realistic soil concentrations of 1–10 mg Cu/kg. Therefore, Cu‐based nanofungicides or fertilizers may have the potential to negatively impact soil mesofauna and other soil organisms when applied as an alternative to conventional Cu formulations. Future risk assessment of CuO‐NP‐ and Cu‐based nanoformulations in soils should consider the role of redox‐reactive (clay) minerals, especially with a focus on low test concentrations.

Supporting Information

The Supporting information is available on the Wiley Online Library at https://doi.org/10.1002/etc.5440.

Acknowledgments

We would like to thank M. Koelling and S. Pape from MARUM Bremen for X‐ray fluorescence measurements. From the University of Bremen, we would like to thank the following colleagues: V. Koch for Cu pore water measurements, J. Rosenau for providing biomarker materials, C. Vogt for X‐ray diffraction measurements and A. Rother for support in AAS measurements. The first author was funded by a PhD scholarship from Hans‐Böckler‐Stiftung within the framework of the graduate school NanoCompetence (PK041). Open Access funding enabled and organized by Projekt DEAL.

Conflict of Interest

The authors declare no conflicts of interest.

Author Contributions Statement

Jonas Fischer: Conceptualization; Methodology; Formal analysis; Investigation; Writing—original draft; Visualization. Ghanem A. D. Talal: Methodology; Investigation. Laura S. Schnee: Conceptualization; Writing—original draft; Visualization. Patricks V. Otomo: Methodology; Writing—review & editing. Juliane Filser: Writingreview & editing; Supervision.

Data Availability Statement

The raw data were uploaded on the repository Pangaea (www.pangaea.de) under the title “Physiological, molecular and behavioural effects of copper oxide nanoparticles and copper chloride on the springtail Folsomia candida exposed via artificial test soils” and will be available after the data curation by the reposiory. All raw data are available from the corresponding author until then ([email protected]).